|
1 /* |
|
2 * Copyright (c) 2005 Nokia Corporation and/or its subsidiary(-ies). |
|
3 * All rights reserved. |
|
4 * This component and the accompanying materials are made available |
|
5 * under the terms of the License "Eclipse Public License v1.0" |
|
6 * which accompanies this distribution, and is available |
|
7 * at the URL "http://www.eclipse.org/legal/epl-v10.html". |
|
8 * |
|
9 * Initial Contributors: |
|
10 * Nokia Corporation - initial contribution. |
|
11 * |
|
12 * Contributors: |
|
13 * |
|
14 * Description: Misc. libxml2 related utilities. |
|
15 * |
|
16 */ |
|
17 |
|
18 |
|
19 |
|
20 #include <charconv.h> |
|
21 #include <EscapeUtils.h> |
|
22 #include <utf.h> |
|
23 |
|
24 #include "CleanupLibXml2.h" |
|
25 #include "LeakTracker.h" |
|
26 #include "Logger.h" |
|
27 #include "XmlEncoding.h" |
|
28 #include "XmlEntity.h" |
|
29 #include "XmlUtils.h" |
|
30 |
|
31 // Local data-types |
|
32 struct ErrorContext |
|
33 { |
|
34 TInt error; |
|
35 TBool retry; |
|
36 }; |
|
37 |
|
38 // Static variables |
|
39 static CXmlEntity* sXmlEntity = NULL; |
|
40 |
|
41 |
|
42 // ----------------------------------------------------------------------------- |
|
43 // EntityResolverL |
|
44 // |
|
45 // A local function to resolve entities found in the buffer being parsed. |
|
46 // ----------------------------------------------------------------------------- |
|
47 // |
|
48 static xmlEntityPtr EntityResolverL(void* /*aContext*/, const xmlChar *aName) |
|
49 { |
|
50 xmlEntityPtr entity = NULL; |
|
51 |
|
52 // Use the standard resolver. |
|
53 entity = xmlGetPredefinedEntity(aName); |
|
54 |
|
55 // Resolve other entities too. |
|
56 if ((sXmlEntity != NULL) && (entity == NULL)) |
|
57 { |
|
58 entity = const_cast<xmlEntityPtr>(sXmlEntity->ResolveL(aName)); |
|
59 } |
|
60 |
|
61 return entity; |
|
62 } |
|
63 |
|
64 |
|
65 // ----------------------------------------------------------------------------- |
|
66 // StructuredErrorReporter |
|
67 // |
|
68 // A local function to collect errors found while parsing the buffer. |
|
69 // See http://www.xmlsoft.org/html/libxml-xmlerror.html. |
|
70 // ----------------------------------------------------------------------------- |
|
71 // |
|
72 static void StructuredErrorReporter(void* aContext, xmlErrorPtr error) |
|
73 { |
|
74 _LIT(KErr, "error"); |
|
75 |
|
76 TBool ignore = EFalse; |
|
77 TPtrC type(KErr); |
|
78 ErrorContext* context = static_cast<ErrorContext*>(aContext); |
|
79 |
|
80 if (context != NULL) |
|
81 { |
|
82 context->error = error->code; |
|
83 |
|
84 switch (error->code) |
|
85 { |
|
86 case XML_ERR_UNSUPPORTED_ENCODING: // 32 |
|
87 // Try to reparse the buffer with another char-encoding. |
|
88 context->retry = ETrue; |
|
89 ignore = ETrue; |
|
90 break; |
|
91 |
|
92 case XML_ERR_UNDECLARED_ENTITY: // 26 |
|
93 case XML_WAR_UNDECLARED_ENTITY: // 27 |
|
94 type.Set(_L("warning")); |
|
95 break; |
|
96 |
|
97 case XML_ERR_INTERNAL_ERROR: // 1 |
|
98 case XML_ERR_DOCUMENT_EMPTY: // 4 |
|
99 case XML_ERR_DOCUMENT_END: // 5 |
|
100 case XML_ERR_INVALID_CHAR: // 9 |
|
101 case XML_ERR_INVALID_DEC_CHARREF: // 7 |
|
102 case XML_ERR_ENTITYREF_SEMICOL_MISSING: // 23 |
|
103 case XML_NS_ERR_QNAME: // 38 |
|
104 case XML_ERR_LT_IN_ATTRIBUTE: // 39 |
|
105 case XML_ERR_ATTRIBUTE_NOT_STARTED: // 41 |
|
106 case XML_ERR_ATTRIBUTE_WITHOUT_VALUE: // 42 |
|
107 case XML_ERR_ATTRIBUTE_REDEFINED: // 43 |
|
108 case XML_ERR_LITERAL_NOT_STARTED: // 46 |
|
109 case XML_ERR_PI_NOT_STARTED: // 47 |
|
110 case XML_ERR_PI_NOT_FINISHED: // 61 |
|
111 case XML_ERR_DOCTYPE_NOT_FINISHED: // 62 |
|
112 case XML_ERR_MISPLACED_CDATA_END: // 63 |
|
113 case XML_ERR_RESERVED_XML_NAME: // 64 |
|
114 case XML_ERR_SPACE_REQUIRED: // 65 |
|
115 case XML_ERR_NAME_REQUIRED: // 68 |
|
116 case XML_ERR_URI_REQUIRED: // 70 |
|
117 case XML_ERR_GT_REQUIRED: // 73 |
|
118 case XML_ERR_LTSLASH_REQUIRED: // 74 |
|
119 case XML_ERR_HYPHEN_IN_COMMENT: // 80 |
|
120 case XML_ERR_TAG_NAME_MISMATCH: // 76 |
|
121 case XML_ERR_TAG_NOT_FINISHED: // 77 |
|
122 case XML_NS_ERR_UNDEFINED_NAMESPACE: // 201 |
|
123 // Just log these as errors and continue if possible. |
|
124 break; |
|
125 |
|
126 default: |
|
127 break; |
|
128 } |
|
129 } |
|
130 |
|
131 if (ignore == EFalse) |
|
132 { |
|
133 FEED_LOG2(_L("Feeds"), _L("Feeds.log"), |
|
134 EFileLoggingModeAppend, _L("FEEDS: parser %S!!! Code:%d"), &type, error->code); |
|
135 } |
|
136 } |
|
137 |
|
138 |
|
139 // ----------------------------------------------------------------------------- |
|
140 // CXmlUtils::NewL |
|
141 // |
|
142 // Two-phased constructor. |
|
143 // ----------------------------------------------------------------------------- |
|
144 // |
|
145 CXmlUtils* CXmlUtils::NewL() |
|
146 { |
|
147 CXmlUtils* self = new (ELeave) CXmlUtils(); |
|
148 |
|
149 CleanupStack::PushL(self); |
|
150 self->ConstructL(); |
|
151 CleanupStack::Pop(); |
|
152 |
|
153 return self; |
|
154 } |
|
155 |
|
156 |
|
157 // ----------------------------------------------------------------------------- |
|
158 // CXmlUtils::CXmlUtils |
|
159 // |
|
160 // C++ default constructor. |
|
161 // ----------------------------------------------------------------------------- |
|
162 // |
|
163 CXmlUtils::CXmlUtils(): |
|
164 iLeakTracker(CLeakTracker::ELibXml2) |
|
165 { |
|
166 } |
|
167 |
|
168 |
|
169 // ----------------------------------------------------------------------------- |
|
170 // CXmlEncoding::ConstructL |
|
171 // |
|
172 // Symbian 2nd phase constructor can leave. |
|
173 // ----------------------------------------------------------------------------- |
|
174 // |
|
175 void CXmlUtils::ConstructL() |
|
176 { |
|
177 // will call XmlEngine::XmlEngineAttachL() |
|
178 iImpl.OpenL(); |
|
179 |
|
180 // Create the XmlEncoding singleton. |
|
181 iXmlEncoding = CXmlEncoding::NewL(); |
|
182 |
|
183 // Create the XmlEntity singleton. |
|
184 iXmlEntity = CXmlEntity::NewL(); |
|
185 } |
|
186 |
|
187 |
|
188 // ----------------------------------------------------------------------------- |
|
189 // CXmlUtils::~CXmlUtils |
|
190 // |
|
191 // Deconstructor. |
|
192 // ----------------------------------------------------------------------------- |
|
193 // |
|
194 CXmlUtils::~CXmlUtils() |
|
195 { |
|
196 delete iXmlEncoding; |
|
197 delete iXmlEntity; |
|
198 |
|
199 // will call XmlEngine::XmlEngineCleanup() |
|
200 iImpl.Close(); |
|
201 } |
|
202 |
|
203 |
|
204 // ----------------------------------------------------------------------------- |
|
205 // CXmlUtils::ParseBufferL |
|
206 // |
|
207 // Returns a libxml2 document from the provided buffer. |
|
208 // ----------------------------------------------------------------------------- |
|
209 // |
|
210 RXmlEngDocument CXmlUtils::ParseBufferL(const TDesC8& aBuffer, const TDesC& aCharSet) const |
|
211 { |
|
212 xmlDocPtr document = NULL; |
|
213 xmlParserCtxtPtr contxt = NULL; |
|
214 ErrorContext errorContext; |
|
215 HBufC8* utf8Buffer = NULL; |
|
216 const xmlChar* parseBuffer = NULL; |
|
217 TInt parseLength = 0; |
|
218 TUint encoding; |
|
219 RXmlEngDocument doc; |
|
220 |
|
221 if (aBuffer.Length() == 0) |
|
222 { |
|
223 User::Leave(KErrArgument); |
|
224 } |
|
225 |
|
226 // Prepare the parser's context. |
|
227 User::LeaveIfNull(contxt = xmlNewParserCtxt()); |
|
228 |
|
229 sXmlEntity = iXmlEntity; |
|
230 contxt->sax->getEntity = EntityResolverL; |
|
231 CleanupStack::PushL(TCleanupItem(&CleanupParseBuffer, contxt)); |
|
232 |
|
233 // Set the error callback |
|
234 xmlSetStructuredErrorFunc(&errorContext, StructuredErrorReporter); |
|
235 |
|
236 // If need be convert the buffer to utf8. |
|
237 if (iXmlEncoding->DetermineCharEncodingL(aBuffer, aCharSet, encoding)) |
|
238 { |
|
239 if (encoding != KCharacterSetIdentifierUtf8) |
|
240 { |
|
241 utf8Buffer = iXmlEncoding->ConvertToUtf8L(encoding, aBuffer); |
|
242 CleanupStack::PushL(utf8Buffer); |
|
243 |
|
244 // Skip anything before the xml prolog. |
|
245 parseBuffer = utf8Buffer->Ptr(); |
|
246 parseLength = utf8Buffer->Length(); |
|
247 |
|
248 SkipCharsBeforeXmlProlog(&parseBuffer, parseLength); |
|
249 } |
|
250 } |
|
251 |
|
252 // Otherwise its already utf8 so use the provided buffer. |
|
253 if (utf8Buffer == NULL) |
|
254 { |
|
255 // Skip anything before the xml prolog. |
|
256 parseBuffer = aBuffer.Ptr(); |
|
257 parseLength = aBuffer.Length(); |
|
258 |
|
259 SkipCharsBeforeXmlProlog(&parseBuffer, parseLength); |
|
260 } |
|
261 |
|
262 // Parse the document. |
|
263 errorContext.error = XML_ERR_OK; |
|
264 errorContext.retry = EFalse; |
|
265 |
|
266 document = xmlSAXParseMemory(contxt->sax, (char*) parseBuffer, parseLength, FALSE); |
|
267 |
|
268 if (document == NULL) |
|
269 { |
|
270 User::Leave(KErrCorrupt); |
|
271 } |
|
272 |
|
273 if (utf8Buffer != NULL) |
|
274 { |
|
275 CleanupStack::PopAndDestroy(utf8Buffer); |
|
276 } |
|
277 |
|
278 CleanupStack::PopAndDestroy(/*contxt*/); |
|
279 |
|
280 // Wrap and return the document. |
|
281 CleanupStack::PushL(document); |
|
282 doc.OpenL( (RXmlEngDOMImplementation&)iImpl, document ); |
|
283 CleanupStack::Pop(/*document*/); |
|
284 |
|
285 return doc; |
|
286 } |
|
287 |
|
288 |
|
289 // ----------------------------------------------------------------------------- |
|
290 // CXmlUtils::CleanupParseBuffer |
|
291 // |
|
292 // The cleanup method for ParseBufferL |
|
293 // ----------------------------------------------------------------------------- |
|
294 // |
|
295 void CXmlUtils::CleanupParseBuffer(TAny *aPtr) |
|
296 { |
|
297 xmlResetLastError(); |
|
298 xmlSetGenericErrorFunc(NULL, NULL); |
|
299 xmlFreeParserCtxt(static_cast<xmlParserCtxtPtr>(aPtr)); |
|
300 } |
|
301 |
|
302 |
|
303 // ----------------------------------------------------------------------------- |
|
304 // CXmlUtils::SkipCharsBeforeXmlProlog |
|
305 // |
|
306 // Skip any chars before the XML-prolog. |
|
307 // ----------------------------------------------------------------------------- |
|
308 // |
|
309 void CXmlUtils::SkipCharsBeforeXmlProlog(const TUint8** aString, TInt& aLen) const |
|
310 { |
|
311 _LIT8(KStart, "<?xml"); |
|
312 |
|
313 const xmlChar* origStr = *aString; |
|
314 TInt origLen = aLen; |
|
315 |
|
316 while (aLen > 0) |
|
317 { |
|
318 if ((aLen >= 5) && (xmlStrncasecmp(*aString, KStart().Ptr(), 5) == 0)) |
|
319 { |
|
320 break; |
|
321 } |
|
322 else |
|
323 { |
|
324 (*aString)++; |
|
325 aLen--; |
|
326 } |
|
327 } |
|
328 |
|
329 // If the xml prolog is missing then do nothing. |
|
330 if (aLen == 0) |
|
331 { |
|
332 *aString = origStr; |
|
333 aLen = origLen; |
|
334 } |
|
335 } |
|
336 |
|
337 |
|
338 // ----------------------------------------------------------------------------- |
|
339 // CXmlUtils::CleanupUrlL |
|
340 // |
|
341 // Resolves any entities and escaped chars in the given url. |
|
342 // ----------------------------------------------------------------------------- |
|
343 // |
|
344 void CXmlUtils::CleanupUrlL(TDes& aUrl) const |
|
345 { |
|
346 HBufC* temp = NULL; |
|
347 |
|
348 // Resolve entities. |
|
349 (void) ResolveEntitiesL(aUrl); |
|
350 |
|
351 // Unescape the url. |
|
352 temp = EscapeUtils::EscapeDecodeL(aUrl); |
|
353 CleanupStack::PushL(temp); |
|
354 |
|
355 // Trim leading and trailing whitespace. |
|
356 temp->Des().Trim(); |
|
357 |
|
358 // Copy temp back to aUrl. This is always safe because temp->Des().Length() |
|
359 // is always <= aUrl.Length(). |
|
360 aUrl.Copy(*temp); |
|
361 |
|
362 CleanupStack::PopAndDestroy(temp); |
|
363 } |
|
364 |
|
365 |
|
366 // ----------------------------------------------------------------------------- |
|
367 // CXmlUtils::CleanupMarkupL |
|
368 // |
|
369 // Resolves any html entities and removes any markup found in the given descriptor. |
|
370 // ----------------------------------------------------------------------------- |
|
371 // |
|
372 TBool CXmlUtils::CleanupMarkupL(TDes& aBuffer, TInt aNewLineChar) const |
|
373 { |
|
374 TBool modified = EFalse; |
|
375 |
|
376 // Strip all markup. |
|
377 modified |= StripMarkupL(aBuffer, aNewLineChar); |
|
378 |
|
379 // Resolve entities. |
|
380 modified |= ResolveEntitiesL(aBuffer); |
|
381 |
|
382 // Strip the CDATA markers. |
|
383 modified |= StripCDataMarkers(aBuffer); |
|
384 |
|
385 // Strip all markup created after entities were resolved. |
|
386 modified |= StripMarkupL(aBuffer, aNewLineChar); |
|
387 |
|
388 // Strip leading and trailing whitespace. |
|
389 aBuffer.Trim(); |
|
390 |
|
391 return modified; |
|
392 } |
|
393 |
|
394 |
|
395 // ----------------------------------------------------------------------------- |
|
396 // CXmlUtils::ResolveEntitiesL |
|
397 // |
|
398 // Resolves any entities found in aOrig. |
|
399 // ----------------------------------------------------------------------------- |
|
400 // |
|
401 TBool CXmlUtils::ResolveEntitiesL(TDes& aBuffer) const |
|
402 { |
|
403 const TUint KEntityStart = '&'; |
|
404 const TUint KEntityEnd = ';'; |
|
405 |
|
406 TInt entityStart = -1; |
|
407 TInt currentPos = 0; |
|
408 TBool entityFound = EFalse; |
|
409 TBool textHasChanged = EFalse; |
|
410 |
|
411 while (currentPos < aBuffer.Length()) |
|
412 { |
|
413 // If this is the beginning of an entity... |
|
414 if (aBuffer[currentPos] == KEntityStart) |
|
415 { |
|
416 entityStart = currentPos; |
|
417 entityFound = ETrue; |
|
418 currentPos++; |
|
419 } |
|
420 |
|
421 // If this is the end of an entity... |
|
422 else if ((aBuffer[currentPos] == KEntityEnd) && entityFound) |
|
423 { |
|
424 TUint16 ucs2Value; |
|
425 TInt entityLength = currentPos - entityStart + 1; |
|
426 TPtrC entityStr(aBuffer.Mid(entityStart + 1, entityLength - 2)); |
|
427 TBuf16<1> entityCode; |
|
428 |
|
429 // Resolve the entity inline. |
|
430 iXmlEntity->ResolveL(entityStr, ucs2Value); |
|
431 entityCode.SetLength(1); |
|
432 entityCode[0] = ucs2Value; |
|
433 |
|
434 aBuffer.Replace(entityStart, entityLength, entityCode); |
|
435 |
|
436 currentPos = entityStart; |
|
437 textHasChanged = ETrue; |
|
438 entityStart = -1; |
|
439 entityFound = EFalse; |
|
440 } |
|
441 |
|
442 // Otherwise just move on. |
|
443 else |
|
444 { |
|
445 currentPos++; |
|
446 } |
|
447 } |
|
448 |
|
449 return textHasChanged; |
|
450 } |
|
451 |
|
452 |
|
453 // ----------------------------------------------------------------------------- |
|
454 // CXmlUtils::StripMarkup |
|
455 // |
|
456 // Remove any markup found in the given descriptor. |
|
457 // ----------------------------------------------------------------------------- |
|
458 // |
|
459 TBool CXmlUtils::StripMarkupL(TDes& aBuffer, TInt aNewLineChar) const |
|
460 { |
|
461 TInt tagStart = -1; |
|
462 TInt currentPos = 0; |
|
463 TBool tagStartFound = EFalse; |
|
464 TBool textHasChanged = EFalse; |
|
465 |
|
466 // TODO: 1) Remove xml comments as well. |
|
467 |
|
468 while (currentPos < aBuffer.Length()) |
|
469 { |
|
470 // It's at the beginning of a tag. |
|
471 if (aBuffer.Mid(currentPos, 1).Compare(_L("<")) == 0) |
|
472 { |
|
473 tagStartFound = ETrue; |
|
474 tagStart = currentPos; |
|
475 currentPos++; |
|
476 } |
|
477 |
|
478 // It's at the end of a tag. |
|
479 else if ((aBuffer.Mid(currentPos,1).Compare(_L(">")) == 0) && |
|
480 tagStartFound) |
|
481 { |
|
482 TBuf<6> tag; |
|
483 TInt tagLength; |
|
484 |
|
485 // Extract enough of the tag to determine if it needs to insert a |
|
486 // newline. |
|
487 tagLength = Min(6, currentPos - tagStart + 1); |
|
488 tag.Copy(aBuffer.Mid(tagStart, tagLength)); |
|
489 tag.LowerCase(); |
|
490 |
|
491 // If it found a <p> or <br> tag insert a newline. |
|
492 // TODO: This fails if the tag includes attributes (i.e. style). |
|
493 if ((aNewLineChar != 0) && |
|
494 ((tag.Compare(_L("<br>")) == 0) || |
|
495 (tag.Compare(_L("<br/>")) == 0) || |
|
496 (tag.Compare(_L("<br />")) == 0) || |
|
497 (tag.Compare(_L("<p>")) == 0))) |
|
498 { |
|
499 // Delete the tag. |
|
500 aBuffer.Delete(tagStart, currentPos - tagStart + 1); |
|
501 |
|
502 // Insert the newline. |
|
503 TBuf<1> chr; |
|
504 _LIT(KFormat,"%c"); |
|
505 |
|
506 chr.Format(KFormat, aNewLineChar); |
|
507 aBuffer.Insert(tagStart, chr); |
|
508 |
|
509 currentPos = tagStart + 1; |
|
510 textHasChanged = ETrue; |
|
511 } |
|
512 |
|
513 // Otherwise just delete the tag. |
|
514 else |
|
515 { |
|
516 aBuffer.Delete(tagStart, currentPos - tagStart + 1); |
|
517 currentPos = tagStart; |
|
518 textHasChanged = ETrue; |
|
519 } |
|
520 |
|
521 tagStart = -1; |
|
522 tagStartFound = EFalse; |
|
523 } |
|
524 |
|
525 // Otherwise just skip to the next char. |
|
526 else |
|
527 { |
|
528 currentPos++; |
|
529 } |
|
530 } |
|
531 |
|
532 if (textHasChanged) |
|
533 { |
|
534 aBuffer.TrimAll(); |
|
535 } |
|
536 |
|
537 return textHasChanged; |
|
538 } |
|
539 |
|
540 |
|
541 // ----------------------------------------------------------------------------- |
|
542 // CXmlUtils::StripCDataMarkers |
|
543 // |
|
544 // Remove any CDATA markers in the given descriptor. |
|
545 // ----------------------------------------------------------------------------- |
|
546 // |
|
547 TBool CXmlUtils::StripCDataMarkers(TDes& aBuffer) const |
|
548 { |
|
549 _LIT(KMarkerBegin, "<![CDATA["); |
|
550 _LIT(KMarkerEnd, "]]>"); |
|
551 |
|
552 const TInt KMarkerBeginLen = KMarkerBegin().Length(); |
|
553 const TInt KMarkerEndLen = KMarkerEnd().Length(); |
|
554 |
|
555 TInt currentPos = 0; |
|
556 TBool markerStartFound = EFalse; |
|
557 TBool textHasChanged = EFalse; |
|
558 |
|
559 while (currentPos < aBuffer.Length()) |
|
560 { |
|
561 // Handle the the beginning of a marker. |
|
562 if ((currentPos <= (aBuffer.Length() - KMarkerBeginLen)) && |
|
563 aBuffer.Mid(currentPos, KMarkerBeginLen).Compare(KMarkerBegin) == 0) |
|
564 { |
|
565 markerStartFound = ETrue; |
|
566 aBuffer.Delete(currentPos, KMarkerBeginLen); |
|
567 textHasChanged = ETrue; |
|
568 } |
|
569 |
|
570 // Handle the the end of a marker. |
|
571 else if (markerStartFound && (currentPos <= (aBuffer.Length() - KMarkerEndLen)) && |
|
572 aBuffer.Mid(currentPos, KMarkerEndLen).Compare(KMarkerEnd) == 0) |
|
573 { |
|
574 markerStartFound = EFalse; |
|
575 aBuffer.Delete(currentPos, KMarkerEndLen); |
|
576 textHasChanged = ETrue; |
|
577 } |
|
578 |
|
579 // Otherwise just skip to the next char. |
|
580 else |
|
581 { |
|
582 currentPos++; |
|
583 } |
|
584 } |
|
585 |
|
586 if (textHasChanged) |
|
587 { |
|
588 aBuffer.TrimAll(); |
|
589 } |
|
590 |
|
591 return textHasChanged; |
|
592 } |
|
593 |
|
594 |
|
595 // ----------------------------------------------------------------------------- |
|
596 // CXmlUtils::ExtractTextL |
|
597 // |
|
598 // Performs a deep extraction of the text children of the given node. The result |
|
599 // is returned as a 16-bit descriptor. The char encoding is always ucs2. |
|
600 // ----------------------------------------------------------------------------- |
|
601 // |
|
602 HBufC* CXmlUtils::ExtractTextL(TXmlEngElement aElement) const |
|
603 { |
|
604 TInt size; |
|
605 HBufC* ucs2Des = NULL; |
|
606 RBuf8 rbuf; |
|
607 |
|
608 if (aElement.IsNull()) |
|
609 { |
|
610 return NULL; |
|
611 } |
|
612 |
|
613 // Perform the deep extraction. |
|
614 size = aElement.InnerXmlL( rbuf ); |
|
615 rbuf.CleanupClosePushL(); |
|
616 |
|
617 // Resolve any remaining entities and strip any CDATA markers. |
|
618 if (size > 0) |
|
619 { |
|
620 ucs2Des = HBufC::NewLC( rbuf.Length() ); |
|
621 TPtr ptr(ucs2Des->Des()); |
|
622 //ptr.Copy( rbuf ); |
|
623 TInt unconverted = CnvUtfConverter::ConvertToUnicodeFromUtf8(ptr,rbuf); |
|
624 while( unconverted>0) |
|
625 { |
|
626 |
|
627 ucs2Des->ReAllocL(rbuf.Length()+unconverted); |
|
628 ptr.Set(ucs2Des->Des()); |
|
629 unconverted = CnvUtfConverter::ConvertToUnicodeFromUtf8(ptr,rbuf); |
|
630 } |
|
631 |
|
632 ResolveEntitiesL(ptr); |
|
633 StripCDataMarkers(ptr); |
|
634 ptr.TrimAll(); |
|
635 |
|
636 CleanupStack::Pop(ucs2Des); |
|
637 } |
|
638 |
|
639 CleanupStack::PopAndDestroy( /*rbuf*/ ); |
|
640 |
|
641 return ucs2Des; |
|
642 } |
|
643 |
|
644 |
|
645 // ----------------------------------------------------------------------------- |
|
646 // CXmlUtils::ExtractSimpleTextL |
|
647 // |
|
648 // Performs a deep extraction of the text children of the given node. The result |
|
649 // is returned as a 16-bit descriptor. The char encoding is always ucs2. This |
|
650 // method is different from ExtractTextL in that it doesn't resolve entities or |
|
651 // do any other clean up. It also has a length param. |
|
652 // ----------------------------------------------------------------------------- |
|
653 // |
|
654 HBufC* CXmlUtils::ExtractSimpleTextL(TXmlEngElement aElement, TInt aMaxLength, |
|
655 TBool aFromEnd) const |
|
656 { |
|
657 xmlChar* xmlStr; |
|
658 HBufC* ucs2Des = NULL; |
|
659 RBuf8 rbuf; |
|
660 |
|
661 if (aElement.IsNull()) |
|
662 { |
|
663 return NULL; |
|
664 } |
|
665 |
|
666 // Extract the text from the element. |
|
667 aElement.WholeTextContentsCopyL( rbuf ); |
|
668 rbuf.CleanupClosePushL(); |
|
669 xmlStr = (xmlChar*) rbuf.Ptr(); |
|
670 |
|
671 // If need be, trim the string. |
|
672 if (aMaxLength > 0) |
|
673 { |
|
674 if (aFromEnd) |
|
675 { |
|
676 if (rbuf.Length() < aMaxLength) |
|
677 { |
|
678 xmlStr = xmlUTF8Strndup(xmlStr, aMaxLength); |
|
679 } |
|
680 else |
|
681 { |
|
682 xmlStr = xmlUTF8Strndup(xmlStr + rbuf.Length() - aMaxLength, aMaxLength); |
|
683 } |
|
684 } |
|
685 else |
|
686 { |
|
687 xmlStr = xmlUTF8Strndup(xmlStr, aMaxLength); |
|
688 } |
|
689 CleanupLibXml2::PushL(xmlStr); |
|
690 } |
|
691 |
|
692 // Convert the value into a TDesC |
|
693 TXmlEngString temp((char*) xmlStr); |
|
694 |
|
695 ucs2Des = temp.AllocL(); |
|
696 ucs2Des->Des().Trim(); |
|
697 |
|
698 if (aMaxLength > 0) |
|
699 { |
|
700 CleanupStack::PopAndDestroy(/*xmlStr*/); |
|
701 } |
|
702 |
|
703 CleanupStack::PopAndDestroy( /*rbuf*/ ); |
|
704 |
|
705 return ucs2Des; |
|
706 } |
|
707 |
|
708 |
|
709 // ----------------------------------------------------------------------------- |
|
710 // CXmlUtils::GetDocumentFirstElement |
|
711 // |
|
712 // Returns the first element in the given document. |
|
713 // ----------------------------------------------------------------------------- |
|
714 // |
|
715 TXmlEngElement CXmlUtils::GetDocumentFirstElement(RXmlEngDocument aDocument) const |
|
716 { |
|
717 const TXmlEngElement KNullElement; |
|
718 TXmlEngElement element; |
|
719 |
|
720 // Get the first child. |
|
721 element = aDocument.DocumentElement(); |
|
722 |
|
723 // Loop through the children looking for the first element. |
|
724 while (element.NotNull()) |
|
725 { |
|
726 if ((element.NodeType() == TXmlEngNode::EElement)) |
|
727 { |
|
728 return element; |
|
729 } |
|
730 |
|
731 element = element.NextSibling().AsElement(); |
|
732 } |
|
733 |
|
734 // Element not found, return a "null" node. |
|
735 return KNullElement; |
|
736 } |
|
737 |
|
738 |
|
739 // ----------------------------------------------------------------------------- |
|
740 // CXmlUtils::GetFirstElementChild |
|
741 // |
|
742 // Returns the first child of the given node. |
|
743 // ----------------------------------------------------------------------------- |
|
744 // |
|
745 TXmlEngElement CXmlUtils::GetFirstElementChild(TXmlEngElement aElement) const |
|
746 { |
|
747 const TXmlEngElement KNullElement; |
|
748 RXmlEngNodeList<TXmlEngElement> children; |
|
749 TXmlEngElement child; |
|
750 |
|
751 // Get the first element -- children only contains TElements... |
|
752 aElement.GetChildElements(children); |
|
753 if (children.HasNext()) |
|
754 { |
|
755 return children.Next(); |
|
756 } |
|
757 |
|
758 return KNullElement; |
|
759 } |
|
760 |
|
761 |
|
762 // ----------------------------------------------------------------------------- |
|
763 // CXmlUtils::GetFirstNamedChild |
|
764 // |
|
765 // Returns the first child of the given node with the provided name. aUtf8Name |
|
766 // MUST be null terminated. |
|
767 // ----------------------------------------------------------------------------- |
|
768 // |
|
769 TXmlEngElement CXmlUtils::GetFirstNamedChild(TXmlEngElement aElement, const TDesC8& aUtf8Name) const |
|
770 { |
|
771 const TXmlEngElement KNullElement; |
|
772 RXmlEngNodeList<TXmlEngElement> children; |
|
773 TXmlEngElement child; |
|
774 |
|
775 // Get the first element -- children only contains TElements... |
|
776 aElement.GetChildElements(children); |
|
777 while (children.HasNext()) |
|
778 { |
|
779 child = children.Next(); |
|
780 |
|
781 if (IsNamed(child, aUtf8Name)) |
|
782 { |
|
783 return child; |
|
784 } |
|
785 } |
|
786 |
|
787 return KNullElement; |
|
788 } |
|
789 |
|
790 |
|
791 // ----------------------------------------------------------------------------- |
|
792 // CXmlUtils::GetNextSiblingElement |
|
793 // |
|
794 // Returns the next sibling of the given node. |
|
795 // ----------------------------------------------------------------------------- |
|
796 // |
|
797 TXmlEngElement CXmlUtils::GetNextSiblingElement(TXmlEngElement aElement) |
|
798 { |
|
799 const TXmlEngElement KNullElement; |
|
800 TXmlEngElement node; |
|
801 |
|
802 if (aElement.NotNull()) |
|
803 { |
|
804 node = aElement.NextSibling().AsElement(); |
|
805 |
|
806 // Loop through the siblings looking for the next element. |
|
807 while (node.NotNull()) |
|
808 { |
|
809 if ((node.NodeType() == TXmlEngNode::EElement)) |
|
810 { |
|
811 return node; |
|
812 } |
|
813 |
|
814 node = node.NextSibling().AsElement(); |
|
815 } |
|
816 } |
|
817 |
|
818 return KNullElement; |
|
819 } |
|
820 |
|
821 |
|
822 // ----------------------------------------------------------------------------- |
|
823 // CXmlUtils::IsNamed |
|
824 // |
|
825 // Returns true if the given node has a name of aName. This method doesn't |
|
826 // compare the namespace. |
|
827 // ----------------------------------------------------------------------------- |
|
828 // |
|
829 TBool CXmlUtils::IsNamed(TXmlEngElement aElement, const TDesC8& aUtf8Name) const |
|
830 { |
|
831 return (xmlStrcasecmp((const xmlChar *) aElement.Name().Ptr(), |
|
832 (const xmlChar *) aUtf8Name.Ptr()) == 0); |
|
833 } |
|
834 |
|
835 |
|
836 // ----------------------------------------------------------------------------- |
|
837 // CXmlUtils::IsNamed |
|
838 // |
|
839 // Returns true if the given node has match the given namespace and name. |
|
840 // ----------------------------------------------------------------------------- |
|
841 // |
|
842 TBool CXmlUtils::IsNamed(TXmlEngElement aElement, const TDesC8& aUtf8NameSpace, |
|
843 const TDesC8& aUtf8Name) const |
|
844 { |
|
845 TPtrC8 ns = aElement.Prefix(); |
|
846 |
|
847 // First ensure the names match. |
|
848 if (!IsNamed(aElement, aUtf8Name)) |
|
849 { |
|
850 return EFalse; |
|
851 } |
|
852 |
|
853 // Return false if one of the namespaces is NULL and the other isn't. |
|
854 if ((ns.Length() != 0 && (aUtf8NameSpace.Length() == 0)) || |
|
855 (ns.Length() == 0 && (aUtf8NameSpace.Length() > 0))) |
|
856 { |
|
857 return EFalse; |
|
858 } |
|
859 |
|
860 // Return false if both are non-null, but don't match. |
|
861 if (ns.Length() != 0 && (aUtf8NameSpace.Length() >= 0) && |
|
862 (xmlStrcasecmp((const xmlChar *) ns.Ptr(), |
|
863 (const xmlChar *) aUtf8NameSpace.Ptr()) != 0)) |
|
864 { |
|
865 return EFalse; |
|
866 } |
|
867 |
|
868 // Its a match. |
|
869 return ETrue; |
|
870 } |
|
871 |
|
872 |
|
873 // ----------------------------------------------------------------------------- |
|
874 // CXmlUtils::AttributeL |
|
875 // |
|
876 // Returns the attribute's value or NULL if the attribute wasn't present. |
|
877 // ----------------------------------------------------------------------------- |
|
878 // |
|
879 HBufC* CXmlUtils::AttributeL(TXmlEngElement aElement, const TDesC8& aAttribute) const |
|
880 { |
|
881 TXmlEngAttr attr; |
|
882 HBufC* value = NULL; |
|
883 RBuf8 rbuf; |
|
884 |
|
885 // Get the value. |
|
886 attr = aElement.AttributeNodeL( aAttribute ); |
|
887 if (attr.NotNull()) |
|
888 { |
|
889 attr.WholeValueCopyL( rbuf ); |
|
890 rbuf.CleanupClosePushL(); |
|
891 value = HBufC::NewL( rbuf.Length() + 32 ); |
|
892 TPtr ptr(value->Des()); |
|
893 TInt unconverted = CnvUtfConverter::ConvertToUnicodeFromUtf8(ptr,rbuf); |
|
894 CleanupStack::PopAndDestroy( /*rbuf*/ ); |
|
895 } |
|
896 return value; |
|
897 } |
|
898 |
|
899 |